OpenAI has struck a major deal to dramatically grow its computing power, signing an expanded agreement with Oracle to support the soaring demands of advanced AI development. As part of the Stargate initiative, OpenAI will lease an additional 4.5 gigawatts of data center capacity in the United States, according to people familiar with the plans.
This move highlights the enormous infrastructure needed to train and run the latest AI models. OpenAI’s products, including its most powerful language and reasoning systems, require vast amounts of electricity and specialized chips to process massive datasets and respond to millions of user queries.
Oracle, known for its enterprise software and cloud services, has become a key partner for OpenAI’s infrastructure ambitions. Their collaboration has steadily deepened as OpenAI expands beyond Microsoft Azure to ensure it has sufficient capacity to scale its AI tools. Renting this much data center power represents one of the largest such commitments in the AI industry to date.
Neither company has publicly disclosed the financial details of the arrangement. However, the size of the deal underscores how competition among AI leaders is driving unprecedented investments in computing resources. Companies like OpenAI, Google, and Anthropic are racing to secure data center space, specialized hardware, and energy contracts to keep up with both enterprise demand and consumer usage.
The Stargate project is part of OpenAI’s long-term strategy to maintain leadership in AI research and product development by building a more diverse and resilient cloud network. By partnering with multiple cloud providers, OpenAI aims to avoid bottlenecks and reduce reliance on any single supplier as it scales its models further.
As AI becomes central to more industries, deals like this illustrate just how much infrastructure innovation underpins the technology millions of people interact with every day.